How to model an institution

Page created by Pedro Scott
 
CONTINUE READING
Theor Soc
DOI 10.1007/s11186-008-9066-0

How to model an institution

John W. Mohr & Harrison C. White

# Springer Science + Business Media B.V. 2008

Abstract Institutions are linkage mechanisms that bridge across three kinds of
social divides—they link micro systems of social interaction to meso (and
macro) levels of organization, they connect the symbolic with the material, and
the agentic with the structural. Two key analytic principles are identified for
empirical research, relationality and duality. These are linked to new research
strategies for the study of institutions that draw on network analytic techniques.
Two hypotheses are suggested. (1) Institutional resilience is directly correlated to
the overall degree of structural linkages that bridge across domains of level,
meaning, and agency. (2) Institutional change is related to over-bridging, defined
as the sustained juxtaposition of multiple styles within the same institutional site.
Case examples are used to test these contentions. Institutional stability is
examined in the case of Indian caste systems and American academic science.
Institutional change is explored in the case of the rise of the early Christian
church and in the origins of rock and roll music.

   One of the most striking developments in recent research on institutions has
been the appearance of a spate of new work by scholars who are using relational
modeling techniques originally developed for the study of social networks to
disentangle the complex logic of institutional processes. Of course, one can argue

J. W. Mohr (*)
Sociology Department, University of California, Santa Barbara, CA 93106, USA
e-mail: mohr@soc.ucsb.edu

H. C. White
Sociology Department, 413 Fayerweather Hall, 1180 Amsterdam Avenue, New York,
NY 10027, USA
e-mail: hcw2@columbia.edu
Theor Soc

that social network analysis has always been about the study of social institutions
and in a certain sense that is surely true.1 But it is also true that there has been a
significant shift over the last decade or so by a number of researchers who have
moved away from a more traditional focus on interactional (e.g., social) networks
towards a modeling strategy of a broader scope that directly tackles the structural
character of institutions by recognizing that what makes an institution work is that
it interpenetrates the social with the cultural.2 Thus an institution links together
different orders and realms of social life, notably the agentic with the structural, the
symbolic with the material, and the micro with the meso and the macro structures
of social organization. Indeed we argue that it is precisely this—the articulation of
relational sub-systems into a structured whole—that constitutes the very essence of
an institution.
    In some cases the distinctive feature of this new research is the application of
network analytic tools to take the measure of cultural systems that underlie broader
institutional processes. Using autobiographical texts as their data, Bearman and
Stovel (2000) map out the network structure of life story narratives as told by
members of the German Nazi party during the 1930s. Smith (2007) uses a similar
model to compare the contested historical narratives of two different ethnic
communities along the Yugoslovian/Italian border. McLean (1998) applies relational
modeling strategies to unpack the cultural logic underlying the presentation of self in
fourteenth and fifteenth century Florentine patronage letters while Sonnett (2004)
uses relational methods to map out the boundaries of musical taste communities in
American society.
    In other work, network tools have been deployed to demonstrate the structural
mappings that link cultural forms with institutional actions. Tilly (1997) used a block
model analysis to show the linkages between the institutional claims various types of
collective actors made upon one another during a series of some 8,000 contentious
gatherings occurring in England from 1758–1834. Breiger and Mohr (2004) relate
the ‘structural equivalence’ concept to logics of institutional practice, and they use a
dual clustering algorithm to map out the relations between cultural identity
categories and types of outreach programs employed by the University of California
in response to an externally imposed prohibition on affirmative action policies in the

1
  Just as an example, early work by the second author (White 1963a) focused on the logic of kinship
systems, building on long traditions of analysis in anthropology, and demonstrated that the institutions of
kinship were derivable through an empirical analysis of the structure of kin ties. In another early
publication on the “Uses of Mathematics in Sociology” the institutional ambitions were expressed clearly,
“What point is there to sociology except as it is able to find and interrelate core properties of, say, a trio
like feudalism (in England in 1200), decentralization in the TVA (today), and political pluralism in France
(nineteenth century)?” (White 1963b, p. 79). Later studies by White carried through on these concerns by
focusing on the changing institutional logic of art worlds, job systems and economic markets (White and
White 1993; White 1970; White 2002).
2
  We focus here on just one style of institutional analysis. It should be said that there are several other
contemporary research projects that are also developing approaches to using formal models to analyze
institutions. The “New Institutionalists” in organizational science is one example (see Powell and
DiMaggio 1991; Greenwood et al. 2008). The use of Boolean algebra and fuzzy set mathematics by
Charles Ragin (1987, 2000) is another important effort.
Theor Soc

mid-1990s. Mische (2007) uses lattices to analyze how Brazilian individual youth
activists representing different ideological perspectives were linked to various social
movement organizations and significant political events during the late 1980s.
   A third cluster of projects has focused on mapping out the institutional
underpinnings of markets, industries and organizational fields. The reconceptualiza-
tion of markets by White (1981) as socially constructed spaces in which producer
firms monitor and orient their behavior toward one another provided the impetus for
a whole new cohort of sociological studies of market dynamics emphasizing the
relational logic of market competitors (White 2002, etc.).
   A somewhat different stream of work has begun to emerge that takes these
models further by using relational methods to measure directly the cultural meaning
systems operating in markets using a variety of sophisticated data collection
techniques. Porac et al. (1995) drew upon a series of interviews with the managing
directors of Scottish knitwear firms to develop a precise relational mapping of rivalry
relations within the industry. Ruef (1999) relied on an innovative textual analysis of
the MEDLINE journal database to generate a relational mapping of organizational
forms in the American health care industry from 1965–1994. Rawlings and
Bourgeois (2004) use an extraordinarily detailed 1945 government dataset to collect
data on credentials offered by American universities that they then use to create a
relational mapping of that institutional space. Kennedy (2005) uses a sophisticated
content analysis of nearly 30,000 (Lexis–Nexis) media stories about the computer
workstation industry that he then applies to test empirically White’s theory of market
monitoring. Many other examples of this style of work could be cited, including a
number of projects that move in very different directions than those referenced here.3
   In this article our goal is to identify some of these broad trends and to unpack the
key assumptions of this new line of institutional research.4 We argue that there are
two central principles that motivate the work—relationality and duality. Both of
these are core theoretical and methodological precepts that have come out of the
social networks tradition but they have been refashioned in this context to embrace a
more holistic sociological vision, one that resonates with the agenda setting
ambitions of the second author’s efforts to reconstruct the theoretical foundations
of contemporary sociology along genuinely relational lines (White 1992, 2008). We
also contend that an effective analysis of institutions needs to take account of three
analytic sites in which linkages occur between different orders of social experience,
those specifying the linkage of agency to structure, of culture to practice and of
linkages across levels of social organization.
   In this article we go on to make two more particular claims about the nature of
social institutions. First, we propose that the stickiness of institutions, those qualities
of institutional life that lead them to be particularly enduring, is directly related to

3
  Mohr (1998) provides a preliminary review of these projects. Ventresca and Mohr (2002) provide a more
detailed review of a number of empirical projects that use these types of methods to analyze markets and
industries.
4
 The article began as a solo piece by White for the UC-SB conference. Mohr brought a fresh perspective
and kept the focus on institutions, as well as introducing new constructs and a great many fresh examples.
Theor Soc

their capacity to bridge these types of divides effectively. In what follows we
illustrate this contention with an example of one institutional form that derives its
resilience from its ability to stretch across levels of social organization and to
combine fruitfully the social organizational properties characterizing different levels
of social interaction under the same institutional roof. Second, we argue that the
undoing of institutional stability, including the possibility for significant institutional
change, may be dependent upon the sustained juxtaposition of multiple styles within
the same institutional site. In a sense then, institutional stability derives from
bridging while instability is the result of over-bridging.
    We demonstrate these assertions by referring to several empirical examples that
are widely divergent in period and ethos. We look at the logic of American academic
science, Indian caste systems, the origins of rock and roll music and the early
Christian church. Our account will necessarily be incomplete because, in spite of all
this good work, there is still quite a bit that we don’t know about how institutions
work and we point to some of these unresolved matters towards the end of our essay.

Defining institutions

To begin, what is an institution? Traditionally social scientists have thought of
institutions as the more enduring or recurrent features of social life. Sometimes the
term is used to describe particular types of conventions or practices for managing
social interactions (the handshake, turn-taking in conversations, etc.), at other times a
broader constellation of activities, values, norms, social structures, and role systems
are understood to define a kind of functional arena or field (the family, higher
education, health care). And yet even at this level of specification there is a wide
divergence in terms of actual definitions. There are two (related) reasons for this.
First, institutions (by their nature) are fundamental, pervasive, and varied features of
social existence. Thus any definition must stretch across a broad and deep expanse
of social life. Second, every attempt to account for or specify such a phenomenon is
immediately implicated in a much broader system of sociological theorizing. To
speak with any specificity of the nature of institutions one must invoke a theory of
actions, persons, social organization, cultural systems and the like and these issues
are still very much in flux in contemporary sociological theory. In what follows we
speak about institutions in the context of a broad theory of social action, worked out
over the course of many years by the second author (White 1963a,b, 1970, 1985,
1991, 1992, 1995, 1997, 2000, 2002, 2008) that locates the concept within a field of
related ideas and expectations about the nature and character of social life and a
structural theory of social science.

Relational models of social institutions

A key assumption of this work is that      social institutions are made up of different
types of interlocking networks. This       includes social networks that link actors
together into various social relations      and role systems. Network analysis has
traditionally focused here, and there      is a long and vibrant scientific tradition
Theor Soc

demonstrating the implications of networks for social organizational processes.5 But
a complete specification involves more than this. Once one begins to unpack the
social phenomenology of a network it becomes necessary to change the way one
thinks about individuals, action, culture, and social process more generally.
    White (1992, 2008) provides extensive theoretical specification that begins not
with individual embodied persons, but rather with identities that can be generalized
to include any source of action. A firm, a community, a crowd, oneself on the tennis
court, encounters of strangers on a sidewalk, each may be identities. Identities are
triggered out of events—which is to say, relational situations that emerge from the
flow of social events, switching from one network situation into another, that
produce a need for control over uncertainty and thus a control over situated actions
and fellow identities. Lives are organized around this flowing movement from one
network situation into another and these situations are themselves phenomenological
sites that include certain other identities (in particular networks) organized around
specific domains of activity (hence they are network/domains or simply netdoms).
Active participation (control efforts) in these situations depends upon both the
network logic and the shared conceptual understanding of what is going on in the
netdom. The shared understanding (localized meaning) is produced through
the interactions (and especially the switching back and forth across these
interactional sites) and embodied in a shared repertoire of netdom stories that
provide the basis for a common understanding of what is going on in the moment.
    The situated meanings, the shared stories, the story plots that give heuristic
structure to shared narratives, and the systems of values that provide a core
categorical substrate for this process, all of these are cultural (symbolic) phenomena
and they too are ordered in relational systems (Saussure 1959). Thus, in addition to
social networks, institutional life is organized around cultural networks, relational
structures that link meanings, values, stories, and rhetorics together into various
structured configurations. An institutional analysis needs to attend to both of these
types of structures, and thus to systems of discourse and systems of social interaction
and to the linkages that tie them together.
    A key assumption of a relational approach is that it is the relations (and the
patterns of relations) that matter. Individual persons, agents, values, or story
elements take on identities (and meanings) in and through their relatedness to others.
It is not the inherent attributes or properties of objects that define them, but rather
their relational location within a field of relations. Formal models of network
relations are geared toward identifying structure preserving (homomorphic)
reductions that allow simpler, core structures to be identified.

Duality and institutional orders

While a focus on relationality has long been the hallmark of formal network
analysis, a second core methodological/theoretical principle emphasizing duality

5
 The most important advance in sociology in the half-century since Parsons’s first book may be
operational accountings of networks. For general surveys of the state of the art see Freeman et al. (1989),
Watts (2003), and Wellman and Berkowitz (1988).
Theor Soc

relations has more recently come to be embraced by network modelers. What is a
structural duality? It is a relationship that inheres within and between two classes of
social phenomena such that the structural ordering of one is constituted by and
through the structural ordering of the other. White (1963a) began to develop formal
approaches for mapping duality relations in his early work on kinship (the duality of
matrilines and patrilines) and it was carried forward into his modeling of vacancy
chains (1970; as the duality of persons and jobs) and early work on structural
equivalence (in the duality of individuals and relations; Lorrain and White 1971).
   An early essay by Breiger (1974) contributed a core formulation and extended the
focus. Breiger took off from an insight in Simmel (1955) emphasizing the duality of
individuals and groups. Simmel’s insight was to see that individuals are largely
defined by the social groups that they are members of but, at the same time (and
dually), social groups are defined by the individuals who are included as members.
With this as a starting point, Breiger developed a set of formal strategies for mapping
out the structural implications of dually ordered social systems and those ideas have
been carried forward vigorously over the last decade and a half (Breiger and Pattison
1986; Boyd 1991; Freeman and White 1993; Duquenne 1995, 1996, 1999;
Houseman and White 1998; Pattison and Breiger 2002).
   The early work of White and Breiger as well as many of the more recent
extensions of these ideas by others has focused on duality as an articulation
mechanism that links one order of social structure into another order (individuals and
relations, individuals and roles, individuals and groups, etc.). In this sense, duality is
a relational device for linking one level of social organization into another level such
that the structural logic of one is revealed to be constitutive of the structural logic of
the other (and vice versa). But a key insight of a broader relational theory of
institutions is that duality is also a property that extends to linkages among different
kinds of social orders. Thus, social structures can be dually articulated with cultural
structures (and vice versa). Thomas Schweizer (1993) was an early proponent of this
insight, and using data from an urban neighborhood in French Polynesia, a group of
hunter–gatherers in Zaire, and a peasant community in Java, he demonstrated the
power of this analytic strategy by modeling the relationship that inheres between the
status logic of material possessions and the social ranking of individuals. In this
work, the status meaning of possessions is interpreted by understanding which
members of the community possess which types of goods. Simultaneously (and
dually), Schweizer assessed the ranking of individual community members in the
social order by observing which material goods they possess.
   Following on this work a number of scholars began using formal methods for
mapping duality relations that link social and cultural components of institutional
systems. Mohr and Duquenne (1997) looked at how cultural categories of the poor
were dually mapped with the social structure of treatment practices by progressive
era poverty relief organizations. Mische and Pattison (2000) analyzed how the social
organizations of Brazilian youth were dually ordered with cultural conceptions of
change. Breiger (2000) studied the dual social relations of Supreme Court justices
and the cultural categories of supreme court decisions. Bearman and Parigi (2004)
map the duality between patterns in social structure and types of conversation topics.
Yeung (2005) analyzes the duality of social relationships (in commune communities)
with subjective (affective) meanings about the character and quality of the
Theor Soc

interpersonal relationships.6 In each case, the analytic task is to illuminate how one
order of social phenomena is relationally linked to a different order of social
experience in such a fashion that their structural inter-dependency is mapped. In the
terms of Giddens (1979), the task is to substitute duality for dualism.

Three key theoretical sites of institutional analysis

We now move more specifically into the question of what constitutes an institution
and how an approach emphasizing relationality and duality can be employed to
model institutional processes. There are (at least) three critical issues that have to be
addressed. These are core conceptual matters that run through the history of social
theory. All have to do with the types of linkages that occur across different orders of
social phenomena. They concern the problems of agency, meaning, and level. Our
goal is to recast these issues in light of the relational approach to institutional
analysis that we are describing here.
1. The duality of styles and institutions:
    At least since Durkheim sociologists have pondered the question of how and in what
sense institutions exist in any independent fashion given that it is individual agents who
act and, through their actions, give life to social institutions. In social theory, this is
often thought of as the problem of the duality of agency and structure (Giddens 1979).
We think of this in terms of the duality that inheres between institutions and what we
call styles. Start with the concept of style. Styles can be thought of as a fundamental
specification of how individual agents live their lives through an ongoing process of
combining understandings of situations with sets of practices arrayed across lives
embedded within social networks (netdoms). Styles are thus an essential element for
social action. More formally stated, a style is a patterning in profile and sequence of
socio-cultural processes across some network population (White 1992, Ch. 5). Such
regularities just reflect types of orderliness and of contingency being perceived and
enacted across that network population. It is through the enactment by agents of
styles that institutions come to have material existence.
    But there is duality here. Styles are not idiosyncratic constructions isolated from
social context. It is through institutions that styles come to be enacted. This is
because a style can be understood only when viewed against the background of an
institution that arrays and disciplines actors’ perceptions of the social organization of
values in which they are embedded. Actions may be erratic and zany under the usual
pressures of contingency and chance from physical as well as social context, and by
the flukes of maneuver for advantage. But the style that drives those actions, by
invoking a fixed set of values correlated with a network in institutional accountings,
it is this style that induces perceptions of regularity. The actors are thus constructing
the social through, and thus because of, the institution that we can read from their

6
  This conceptualization of duality and its relationship to research coming from formal network analysis
traditions is analogous (both theoretically and methodologically) to several other sociological projects,
most obviously with Pierre Bourdieu’s efforts to use correspondence analysis as a way to measure the
duality of social and cultural structures. See Breiger (2000) for an especially useful discussion.
Theor Soc

behavior. A network population conceives of behaving and misbehaving, evinces a
style, in terms of the institution, that itself is the carapace left from the enactments of
style.7
   For example, each debate evolving among factions within research communities
over theories in a science is shaped in terms of the institution immanent in its social
organization, and conversely. Changes in invoking and expression of particular
values may seem to be triggered by environment or by group and individual whim,
but change in their structuring as a package across networks is difficult because style
is constrained by institution.
   As introductory example consider a case study of scientific research. Susan
Cozzens (1989) dissects a controversy in neural pharmacology, which amounts to
the social construction of a “discovery.” Expressions of the values of honesty and of
originality are being negotiated through various agencies. One can read an institution
from this particular social formation for a sub-discipline of science. Multiple
networks—of collaboration, of training and sponsorship, of gossip, of friendship and
the like—are sources as well as products of rankings in this case of Cozzens. These
rankings are both particular, as scientific pecking orders of investigators (Chase
1980) in specific subfields surrounding neural pharmacology, and also they are more
general as status-layering, personal and institutional, in the larger scientific
population.
   This case also exemplifies how style spans across institutions in a complex
network population. Mass-media news conferences were one sort of agency being
invoked. Another was brokering and conciliating by other senior scientists. All meld
into activations and adjustments of network and standing among the four different
working groups vying for recognition as co-discoverers.8 Dual to all this is the
semantic negotiation of concepts and perceptions, in struggles over whether there is
an “it” to be discovered, and over bounds and shape for this “it”—the “opiate
receptor.”
   Our point is to note the interaction between style and a cross-sectional pattern of
socio-cultural organization that we denote by institution. In this article, our argument
is that style can best be understood when viewed against the background of an
institution that arrays and disciplines actors’ perceptions of the social organization of
values in which they are embedded. The actors indeed are constructing this
organization through and thus because of the institution that we can read from their
behavior.
2. The duality of the social and the cultural:
  A second key analytic site for a specification of a relational theory of institutions
concerns the problem of meaning and the way in which the meaningful (interpretive)
experience of everyday existence is dually linked to the physical (embodied)
experience of agents linked through social structures. Meaning is essential to how
agents negotiate everyday life. At the level of individual agents, what matters are the

7
  Many of these perceptions and enactments within the population occur in verbal discourse, and in a
follow-on paper White specifies some interrelation between style and micro-dynamics of language.
8
 It is clear that many other individuals and organizations and scientific interest groups also push claims of—
and upon—particular research groups like these.
Theor Soc

ways that interactions in netdoms are interpretable and how switching back and forth
across netdoms generates gradients of understanding (White 1995). At this level of
social existence interpretive systems are organized as stories that are told to oneself
and shared with others about the immediacy of events, actions, and agents that are
known, witnessed, and participated in. Within any given network, meanings are
shared through collective participation in repertoires of stories that serve to give an
interpretive face to the lived experience of interactions. Stories are themselves
constructed out of a patterned relation of meanings (these can be modeled as
networks, see Bearman and Stovel 2000; and Smith 2007) and they also operate in
sets, so there are relational semiotic systems both within and between stories. The
duality that links these semiotic systems back to the embodied practical reality of
social interactions establishes the foundation for the hermeneutics of everyday life.
   The same is true of institutions except that with the institution we move up a level
of social organization and also up a level of abstraction. Interpretive frames at the
institutional level are not immediate and specific; they are general and abstract; and
they are collective rhetorics that are known, shared, and embraced as orienting
devices for the construction, consumption, and bundling of stories across different
netdoms. By invoking a fixed set of stories and institutional arrangement thereof,
rhetoric induces a perception of regularity for participants within an institutional
realm.
   Values are particularly critical for institutional rhetorics. Values are organized in
relational (semiotic) systems that provide anchors for the interpretative under-
standings of what is going on within an institutional realm and provide a moral
coding for those understandings that helps orient participants towards what is
important, what is appropriate, and what will be valued by others who share this
institutional space.
   The first point to note about this is that values code from social network. Values
can be read out of patterns among affective ties, or from patterns in chains of agency,
instead of turning to style to parse meanings from stories and speeches. This is as
true when participants do the reading as when observers do it. That is illustrated in
the next section by the coding of purity stacking from patterns of ritual interchange
among caste, and in the last section by the coding of outreach mission from the
expanding-tree networks of an early church.
   A value has some connotation of ranking, commutation, degree, being in or
defining dimension. This has some cognitive basis (De Soto 1961; De Soto and
Albrecht 1968; Shepard 1984). Such “dimension” partakes of ordering within social
spaces of networks as well as of ordering in semantic space, and yet it also must
ground institutional system in mundane space–time of biophysical perception
(Gibson 1950, 1979).
   Some values, as in asceticism, may appear to be removed entirely from social
interaction. Values can have their own cultural pecking orders, with sacredness
enveloping the high ones, but equally striking is their multiplicity of format, from
commandments over to mere tones within scripts of interaction. Earlier theoretical
perspectives were typically focused on values in themselves and in abstract form.
Talcott Parsons, in his efforts toward a general theory of social action, pushed for
transposability and separability of values so that the ultimate ends of different
means–ends chains could constitute an abstract system of values. Thus for Parsons
Theor Soc

ultimate ends can be cultural facets of coherent social action across a society.
Parsons seeks to strip values out of styles tied to social organization.
   This is a far cry from values embedding with networks in institutions, as can be
found in Erving Goffman (1955, 1963, 1967, 1971, 1974) who identified an
institution in everyday life among strangers. There is a “backstage” and a front stage
where reality and valuation of even the simplest social acts must be negotiated
afresh, endlessly.9 Others, such as DiMaggio 1987; Habermas 1975; March and
Olsen 1976, 1989; and Merton 1968, also push for explanation of meaning and value
being in interaction with tangible organization. These others typically work, like the
present article, with middle-range orders.
   Within institutions values operate in sets. One can no more invoke a value alone
than clap with one hand. Lessons in white lies of tact and adaptability go along with
lectures on honesty as a natural package for ordinary living (White 1991). Or take
values in working science: originality (a variant of scope) is a preeminent value, but
so also is its complement and obverse in scientists’ thinking and talk, the value of
truth (a variant of purity). As here, the values in a set need not be seen as hostile but
instead may be viewed as natural complements like hammer and chisel.10 From one
population, tribe, town, or factory, to another, even nearby, one may code a
somewhat different set of values, and of course distinct networks. But the main shift
will be in relative frequencies of use of a given set, and so it occurs in how given
values are mapped onto sorts of interaction across networks.
   The meanings of a package of values are to be inferred as much from the social
architecture as vice versa. Networks can bring and indicate change as well as nest
actors into shells that block action (Burt 1990). Change may be only a diffusion of
small changes within the envelope of a style with values fixed. But the changes can
be larger.
   Return to the Cozzens example. There values do not operate as universal guides
that have been transposed to this scene. The ordinary language terms of “priority”
and “originality” dissolve in confusion amidst the bewildering struggles and claims,
which manufacture, as convenient, new criteria of connection and information. To
transpose—as a term in an algebraic equation or in a sentence—means to change to
a new location without altering the value, whereas “priority” transposed to the opiate
receptor case is indeed altered.
   A second illustration comes from a case study of a social movement. Civil Rights
values, seen to include purity and effectiveness, were woven into the social fabric of
Movement organization. In the words of McAdam (1988, 237): “Activism depends
on more than idealism…. Formal organizations or informal social networks must
also exist to structure and sustain collective action. The volunteers were not
appreciably more committed to Freedom Summer than the no-shows”—among the

9
  For a cogent overview of the phenomenological position, which situates Goffman with Sachs and
Garfinkel, see Rawls (1989). For incisive appreciation and critique of Goffman see Burns 1991. Hallett
and Ventresca (2006) provide a thoughtful discussion of how to reconcile these types of symbolic
interaction approaches with a contemporary understanding of institutions.
10
   Until some proper calculus has been developed for analysis of discourse—perhaps along some such
lines as Abell (1987) or Chafe (1994)—metaphor and intuition will have to serve. Consult Gibson (2003)
for important recent steps.
Theor Soc

college students initially recruited; so that it was position in mobilization networks
that tipped who actually came, rather than intensity of individual beliefs in Civil
Rights values. We see how institution is the basis for the correlation of style to
overall social arrangement.
3. Dualities among nested levels of social life:
    The third analytic site for a relational theory of institutions has to do with the sort
of duality that links one level of social experience into another. In the history of
social theory the problem of levels has been confronted in many ways. One thinks of
the theoretical conundrums associated with the duality of self and other, of the
individual and the group, of local social organization and field level organization, or
more generally, the linkage from more micro to more meso to more macro levels of
social experience. As suggested earlier, it was the problem of articulation of one
level of social organization into another that originally led to Breiger’s operation-
alization of his theory of structural duality. The problem is central for a relational
theory of institutions because the concept of institution is in its essence a meso level
social phenomenon.11 Indeed, as we argue below, it is precisely the ability of a given
institution to bridge effectively across levels of social organization that often
distinguishes the more enduring and robust institutional forms from others that are
more transitory.
    The use of structural duality measures as a way of analyzing these types of
linkages across institutional levels was initially theorized by Fararo and Doreian
(1984). In what they termed “a formal theory of interpenetration” Fararo and
Doreian sought to generalize and extend Breiger’s bipartite graph model to a
tripartite graph solution that enabled them to move beyond the duality of individuals
and groups to incorporate the duality of groups and organizations also (defined as a
series of connected groups).12 But their ambition was more expansive than simply to
enlarge the scope of structural duality metrics, Fararo and Doreian linked this
endeavor back to classic analytic problems in Marx, Mead, and Simmel and
proposed that such an analytic project could provide a way to create formal models
of institutions by structurally linking the analysis of persons, collectivities, and
cultural systems.
    Other scholars have also focused on the articulation across levels as a key
problem for institutional analysis. Indeed, a major focus of the new institutional
approach to organizational analysis has been how to conceptualize the relationship
between individual organizations and what Scott and Meyer (1983) define as
organizational sectors or, in DiMaggio and Powell’s (1983) formulation, the
relationship between organizations and organizational fields.

11
   Indeed, Miller argues that it is a mistake to equate social conventions (such as the handshake) with
institutions. “Social institutions need to be distinguished from less complex social forms such as
conventions, social norms, roles and rituals. The latter are among the constitutive elements of institutions.
Social institutions also need to be distinguished from more complex and more complete social entities
such as societies or cultures, of which any given institution is typically a constitutive element” (2007, p. 4).
12
   Fararo and Doreian (1984) also reference the work of Thomas Wilson (1982), whom they acknowledge
for having a parallel intellectual discovery with Breiger of the bipartite graph concept.
Theor Soc

   Figure 1 provides a visualization for these arguments. The three circles
correspond to different levels of social organization (self, group, field) that are
linked together by structural dualities. But within each circle, other relational
dualities define the space. At the level of the self there is a duality that inheres
between mind and body, of knowing and doing. In a group, there is a duality linking
stories to social networks. At the level of an institutional field, there are systems of
rhetoric and systems of social organization. Our contention is that institutions
subsume these relational sub-systems and the various articulations that link them
together. Consider the following examples.

Institutional endurance: village caste and academic science

A first institution is illustrated in two very different cultural guises, caste and
science. In the first of these guises, change in either institution or style is generally
thought to be precluded, whereas in the other dress the institution is generally
thought to encourage change in both. Meticulous study of a particular system is
required to support description of an institution. For caste we rely on the account of
Mayer (1960) of a field of villages in central India that maps out networks of
behavior and perception. The second illustration is from an utterly different
discourse, academic science as a social formation in the United States, in which
Cozzens’s case earlier is a particular drama.
   Accounts of Indian caste in general (Dumont 1986), emphasize the value placed
on ritual purity and the strict demarcation of bounds for corporateness. A curious
topology underlies and justifies both emphases, within the ongoing networks of
kinship across villages. This topology very much depends on how networks weave
together a corporate structure of purity. The topology can be tagged as a metonymy,

Fig. 1 Nested dualities within an institution
Theor Soc

specifically a synedoche: “the smaller contains the larger.” Later we will transpose
this topology to modern academic science and claim a good fit.
   Purity is operationalized with extreme explicitness in villages. There is a ranked
series of substances whose passage from one to another grouping permits precise
imputation of purity, hence of socio-cultural standing. The groups ranked are local
and constitute a partition of all family units within a given village into specific caste
groupings. As we saw earlier in the analysis of cultural status and social structure in
the case of Schweizer (1993), this is a system that dually orders forms of social
structure and the relational logic of cultural values. Rankings of each caste can be
imputed from observed transactions, especially those on public occasions as in
religious feasts and weddings. Party food, cooked in butter, comes at one extreme,
then ordinary food, then raw food, and on through smoking pipes, drinking water
and on down through garbage and feces.13 A generally agreed linear order of village
castes is supported thereby.
   The village is the site of primary economic activity, farming and artisanry. Caste
matters do not tie directly to most of this ecological activity. Nor do struggles over
ownership, improvement, and change in property, which largely lie outside the scope
of the village.
   Through marriage kinship cuts across villages. The unit of intimacy is the subset
within a village caste who have married into a like subset of an analogous caste in
another village, or rather the many such intermarrying sets across a number of
villages in the region. Call this the subcaste, a network construction hopping across
the region. It is a “sub”-caste because within any particular village the persons in the
given affinal network are but a subset of the several clusters of blood relatives that
make up the whole caste in that village.
   Paradoxically, the subcaste is larger than the caste. The reality is that inheritance
and marriage, the engines of major change, lie within the subcaste and outside the
village. There is both village exogamy and caste endogamy. The corporate reality of
a subcaste interlinked through affinal ties across a whole region, this intimate
corporateness is broader than the only caste unit that is actually embodied, that of the
village. Indeed the subcaste may be comparable to the whole village in size.
   Only the subcaste has explicit organization, a council and like agencies for
regulating caste affairs. Wealth flows along subcaste lines, through marriage and
inheritance, as do innovations, material and other. There are only the thinnest threads
of purity calibration that can be spun out by the Brahman “priests,” themselves
scattered as local caste units in villages.
   Some mobility can take place without disturbing perceptions of purity value. It is
the mobility not of individual persons but of whole subcastes moving to new
villages. They can do so because there are more distinct castes in the region than can
ever be found together on the ground of a particular village. The Brahman argues
primarily in terms of the four broad varnas of their scriptures; so a subcaste new to
the village can appear and argue for a location in purity rank within that village in
the only way that matters, getting the appropriate exchanges going.

13
     Explicit network modeling of such caste interaction data can be found in Marriott (1959, 1968).
Theor Soc

   Looking at Mayer’s photographs of separate caste groups hunkered down in
separate locales at feasts, one can almost supply the story sets used to account for
and tidy up the value-ranking by purity. Not able to be photographed, but equally
central, are the sprawling networks of kinship bonding across other villages that
sustain and reproduce the system. A separate set of value facets is embedded in these
ties among subcaste segments.

American academic science

There is a close parallel with the organization of American academic science. For
village read university, and for village caste-group read department in a university.
Purity becomes prestige, itself a stand-in for degree of ultimate truth. Subcaste
translates into specialty, which is at the heart of science as research, as generator of
originality. With this translation, the statements above on Mayer’s village caste
system can be carried over into the social setup of American academic science.14
   A pecking order for scientific disciplines is operationalized only among
departments within the particular university, physics or mathematics often being
top and sociology near the bottom.15 Degree of scientific purity may be attributed to
this order. Ritual pervades this scene, whereas action and excitement and intimacy
grow along research networks spreading outside the given university. Sprawling
networks of collaboration and intimacy in actual research reach across the nation on
specific subjects. Specialties are concretized in them. For specialty as subcaste read
invisible college (Crane 1972).
   Seen from outside within its university, a given department has meaning and
indeed coherence. The meaning comes within and expresses a hierarchy of purity
and accomplishment, which can vary from one to another university. The hierarchy
is enforced by deference behaviors in committee meetings and luncheon inter-
changes. It is also enforced and expressed in larger and more solemn ceremonials, at
which symbolic capital can be exchanged like any other.16 Contents of interchanges
can be typed analogously to the discriminations from high to low in the material
exchanges among castes within a village, and there is a parallel range of meeting
contexts within a village.
   Seen from its inside, a university department of science, analogous to a village
caste, is a shambles of unrelated specialties. Different collections of specialties will
make up that discipline’s departments in various universities. And universities differ
in their exact menus of departments. All this is parallel to the caste/village/subcaste
formation. A specialty spreading across scores of departments is typically much

14
  It may be that the dissection below carries through as well to the humanities, with whose organization
we are less familiar. Better analogues to the humanities, as well as to professional schools and the like,
may be the ethnic enclaves that crosscut villages and castes in a region like Mayer’s.
15
  This pecking order varies a little among universities. Sociology, for example, is higher at public
universities than in Ivy League schools.
16
   Consider, for example, the Feast of Grades meetings annual in Arts and Sciences Faculties, at which
higher honors for graduating students are negotiated and solemnized. One might speculate that in the
university, as in the village, these local ceremonials absorb energies and distract attention from flows of
main action and resources through corporative networks.
Theor Soc

larger than a given university department—just as a given subcaste spreading into
scores of villages has more members than are in that caste in a given village.17
    Each specialty is the prime world of motivation for its working scientists. And
when recruitment of new (faculty) members to the given department comes up, it is
recruitment largely along invisible-college networks of specialties active in that
department and competing with each other. Just so do marriages and children get
formed through networks within a subcaste quite separate from the other subcastes
that together in the eyes of that village make up the weavers (or the blacksmiths or
whatever) as a single caste entity. One specialty is linked to its departmental
counterparts in other universities differently from the way another specialty is—
another reason why they tend to operate separately and independently within their
own department. Again there is a parallel to Mayer’s descriptions for subcastes.
    Familiarity of the reader with our academic setting can supply detail and
conviction, though excellent case studies like Cozzens’s cited above can help. For
overviews to supplement case studies and personal experiences, see, for example,
Merton 1973, and Mullins 1973. For longitudinal perspective on values, showing
how careers play out in science see Cole and Zuckerman 1979, and Menard 1971.
    Now we can summarize: Because the caste illustration is a vivid one, this first
institution is easy to recognize in other contexts such as modern science. A general
value—call it purity—is here operationalized in the strongest kind of social ordering,
a ranking close to the full linear ordering of the mathematician. But this purity is
confined operationally to a small locality, usually a geographic one defined by
mundane activities. Purity thus is operationalized only for small populations
separately, since linear order can be enforced only through transitivities in chains
of behavior that cannot be monitored in larger groups.
    And yet the local level of social organization is insufficient. What makes this
institutional form robust is precisely the way in which the structural logic of each
level (and the articulations among them) contributes independently to the overall
coherence of these domains of social life. Everyday intuition suggests that corporate
membership should play the role of locality, but in fact subcaste sprawls out across
localities without, however, interfering with behaviors whose patterns are sufficient
to enforce purity among subcastes within a locality. For either science or caste, other
value facets, such as intimacy and wealth, variously accompany the localities and
corporate memberships. What persists and reproduces itself as a robust social
formation, an institution, is some peculiar balancing of interpenetrations capturing
the duality of cultural values and social networks as well as a complex structural
mapping that links the local and the field levels of social organization.18 Throughout
all of this, one finds individual identities enacting styles that trace out the meaningful
articulations of these various dually articulated structures.
    The view of caste described here required a new generation of anthropologists
who were alerted to networks as analytic and phenomenological bases of social

17
     Even the numbers are quite comparable, for sizes of units and subunits and spread of networks.
18
  In this respect, it is interesting to recall that Durkheim (1933) saw professions as important in modern
societies precisely because they provided a mechanism for re-inscribing the kind of moral sensibility that
had traditionally been viable in the local community but at a higher level of social organization.
Theor Soc

topography and perception. They came to see values as byproducts from social
pattern rather than as exogenous cause. It became clear to them that particular values
in a package were tied together and mapped very differently in the sprawl of
corporate networks than in the prestige layering within a locality.
   Sociologists of science have been bringing in this same new network perspective,
and their results can be mapped into the same institution as village caste. So one can
argue for common predictions as to some matters. The styles accompanying this
institution yield a social organizational system that seems peculiarly resistant to
attempts at external control. Anecdotes on caste and science suggest this as does the
institution’s having a balanced and stable yet decentralized skeleton.
   Caste is indeed argued (Ghurye 1961; Hocart 1950) to be the resilient formation
to have survived repeated waves of external invaders. And the structuring of
American academic research science may well be a reaction against the emerging
dominance of American universities by autocrat Presidents in the early 1900s. It is
not particular values taken as abstract symbols that account for this obduracy; it is a
style that embeds value facets operationally into a certain sort of balanced formation
as institutional system. Indeed, our conjecture is that it is precisely the way that
institutions bind these multiple elementary components into a larger system of dually
articulated relational structures that defines the very essence of institutions as
enduring social formations.

Institutional transformation: styles as vehicles of change

All of this then raises the question of how and in what ways change is enabled or
facilitated within an institutional order. The ideology of science holds that change of
formulation is easy, a convergence of individual perceptions and cognitions subject
only to verification of the truth of findings prompting such reformulation. The social
organization of active research in science is in “invisible colleges,” each a cluster of
networks of workers whose themes cohere around a subfield within any recognized
major discipline. In actuality, as studies of invisible colleges show, reformulations,
even minor ones, are matters of contention.
   The main point is that no conceptual, that is to say cultural reformulation, can
take place apart from some change and adjustment in social organization among
those concerned, producers and critics and users. And a major reformulation requires
a wholesale revamping or even replacement of an invisible college. This is no easy
matter. A new paradigm usually turns out, upon examination, to be associated with a
whole new invisible college (Kuhn 1970).
   It is often argued that invisible colleges emerge because of the difficulty of
building within a whole discipline a critical mass for and sympathy toward a new
line of work. The devil here is the existing, possibly ossified Establishment
ensconced within separate departments of the discipline around the country.
Certainly no one department is large enough to mount work on the scale of an
invisible college. But it is not at all clear that a department is a bad place for
generating new ideas; indeed the crossing of sub-disciplines which departments
afford would seem almost a precondition for innovation.
Theor Soc

   The hieratic layering imposed on departments at a given locale may offend and
hurt, or exhilarate, but it has almost nothing to do with the bulk of research, which is
not implicated across departments. The really extensive and possibly pernicious
stratification is that of individuals by reputation cumulated from research networks
of invisible colleges (cf. sociology of science cites earlier). What departments do is
afford long life to invisible colleges by giving them multiple bases that become very
hard to control, much less to change or abolish by any other authority.19 All can still
be well, of course, for innovation and reformulation if leading figures in the invisible
college are pure of heart. In short, ideology reigns.
   Hence change of style, which is also to say change of institution, is difficult. Such
change comes only through messes and fights, and emerges out of chaos. In analytic
terms, change can be construed as superposition of distinct styles, here of separate
invisible colleges, which for a while overlap and interpenetrate around some new
theme or topic. A genuine reformulation, we argue, will prove not to be encapsulable
within either of the existing invisible colleges. The new formulation could take over,
colonize one of them, the other breaking back off. But more commonly, we suspect
from existing studies, the new formulation breaks apart from both antecedents
(which thereby disengage) and sets up on its own.20
   As an example, take the new biophysics after World War II. Or, more recently and
close to home, take the emerging new invisible college of social studies of science,
which descends from an unquiet and temporary merger of Mertonian and Garfinkel
notions and disciples. The list of possibilities is endless.21 But is this just a
peculiarity of this particular modern Western enterprise, science? We argue: no.
Exactly the same line of argument, with obvious changes in parameters of scale and
wealth and timing, should apply for field investigators of the Indian village caste
scene. In fact, details of castes do change much more than the stereotype allows for.
   Within a village, a given caste changes only through influences that come in, via
dowry and inheritance and notions, along networks of subcaste. More broadly and
regionally, a caste makes itself known through the impact of various constituent
subcastes, so that it is their changes, ironically, that count. These changes are not
common and do not occur easily just because of the way subcastes interlock through
village subunits. The record is that major change in subcaste comes through the
same kind of merger and split we conjectured for science.

19
  Ironically, invisible colleges do provide a way, about the only way, to change local styles of
departments, which tend toward an extraordinary degree of ossification when there are no active
subdisciplines within to shake them up.
20
    A reader, Paul DiMaggio, comments: “Interesting…. What about centralized efforts to use universities
rather then departments or networks to implement change by seeding subcaste members in many different
departments. I'm thinking of the Ford Foundation's support for behavioral political science in the 1950s.
It's relatively easy for a centralized agency to convince a university administration that they should take
chair money for a new type of person; and because the subdisciplines are spread out, the cost to any
department of admitting one new member of a previously unrepresented tribe is low, and easy for the
university administration to leverage. Insofar as the networks control the central agency (e.g., NSF), this
won't work, so it may be a special case.”
21
     And of course each would require sustained inquiry with careful definition and measurement.
Theor Soc

Institutional transformations: rock and roll and the early Christian Church

We turn now to a second set of examples of institutional change, a set of examples
that extend our previous comments to illuminate more fully the role of juxtaposed
multiple styles in transforming existing institutional structures. Our argument here is
more of a conjecture, but we suggest that a new institutional system requires an
intermediate period of overlay and melding between two existing ones, with one or
both of these possibly separating out again for subpopulations.22 We test this
conjecture in diverse cases. New institutional systems correspond to radical changes
in style, to innovation, and we search for such innovations in style to test if the
conjecture holds.23
   More precisely, our conjecture is that major innovation in style can emerge only
from the superposition, for a time, of two or more existing styles, with attendant
institutions. The new style can result in any of several stances vis-à-vis the old
styles. There may follow an untangling and rejection between the original two, along
with the persistence of the novel style. It seems fair to characterize this in shorthand
by the aphorism “styles mate to change” (White 1991).
   Another institution, reminiscent of that implied in McAdams’s study of civil
rights mobilization, underlies the two remaining studies of style change. Catholic
Christianity emerged out of encounters between pre-existing Judaic sects, baptismal
and Pharisiac (cf. e.g., Dix 1947; Flusser 1988; Schoeps 1969). Doctrine, its
existence and its change as a structure of values, is inextricably intertwined in each
with its social networks and dimensions of ranks as formed among localities. We
argue, as before, that such styles are so locked in that they can change only through
turbulent overlay one upon another.
   Turbulent overlays were exactly characteristic of the network of cities that
constituted Asia Minor and the Syriac region for these sects. In modern terms we
would refer to the people involved as lower middle class, often artisans, who had
become strangers to peasant life and tribal contexts. For centuries there was a
diaspora of what we today would call a religion among Hebrews across these cities.
Associated on the fringes were God-fearing Gentiles, who did not fully partake of
the behavior proscriptions and prescriptions or the orientation to a single central site
in Jerusalem.
   Consider this scene when several streams of the infant Christian movement were
already established as styles. Their own separate emergences often are too obscure in
the historical record to pursue further. For example, the basic facts of heresy and
schism between “Gnostics,” who were furthest from Judaic concepts, and
“Christians” continue to be debated (e.g., Grant 1966; Robinson 1988).
   Take the Ebionite Church as one stream. This use of the term Ebionite is
anachronistic. The basic point of Schoeps (1969) is that the Ebionite Church of
Transjordan and Arabia, as a fossil form of Jewish Christianity, held in suspension
for several centuries a synthesis of styles different from what we call Christian—and
then recruitment of new populations and social networks by Mohammed amidst a

22
     Abbott (1988) makes a similar argument about the genesis of new professional groups.
23
     Note that in this view fashion is part of style rather than innovation in style.
You can also read